Sunship Audio

Testing Crossover Phase Coherence in Horn Loudspeakers

A crossover network divides the musical signal between drivers, but its job is more demanding than simply setting frequency bands. The electrical filter, driver response, horn geometry, cabinet alignment, and acoustic placement all combine to determine whether adjacent drivers reinforce one another or partially cancel. Testing a crossover network for phase coherence therefore requires measurements of the complete loudspeaker, not just inspection of its capacitors, inductors, and resistors.

In a horn-loaded system, small timing errors can become especially audible around the crossover region. Compression drivers may have different acoustic centers from woofers, while a deep wooden horn introduces its own path length. A passive network can be carefully designed and still require physical alignment and listening verification before the acoustic result is coherent.

The goal is not to force every phase trace into a perfectly flat line. The practical aim is consistent acoustic arrival, smooth summation, stable imaging, and natural energy through the crossover region. Sunship Audio’s design notes provide useful context for understanding how horn geometry, driver choice, and cabinet construction influence that result.

Why Phase Coherence Matters

Phase describes the timing relationship between a waveform and a reference. In a two-way or three-way loudspeaker, the important relationship is the phase difference between adjacent drivers at and around their crossover frequency. If their outputs arrive together, they add constructively. If they arrive out of step, the response develops a dip, uneven directivity, or a blurred spatial image.

Amplitude measurements alone can hide this problem. Two drivers may each show a smooth frequency response while their combined output contains a deep cancellation. The effect can vary with microphone distance and listening angle because phase is linked to acoustic path length and radiation pattern.

A phase-coherent crossover usually produces a clean handoff rather than a detectable change in character. Voices remain stable as they pass between drivers, percussion retains attack, and stereo placement is less sensitive to small head movements. These benefits are particularly valuable in high-sensitivity horn systems, where microdynamic detail makes timing errors easier to notice.

Prepare the Loudspeaker and Measurement

Begin with a stable measurement setup. Place the loudspeaker on a rigid stand or its intended support, aim the horn at the microphone, and use the same reference position for every sweep. Keep the microphone far enough away for the drivers to combine acoustically, while remaining close enough to avoid excessive room reflections. A gated impulse response is usually preferable in a domestic room.

Measure at a moderate level first. Excessive output can introduce compression driver distortion, amplifier clipping, or woofer nonlinearities that contaminate the phase result. Use a calibrated measurement microphone and record the sampling rate, microphone position, amplifier gain, and crossover configuration so that later comparisons remain valid.

Before testing phase, confirm basic electrical safety. Disconnect the amplifier, inspect solder joints, and verify component values against the schematic. A reversed capacitor, incorrect inductor, or loose terminal can imitate a phase problem. For a passive network, an impedance sweep is also valuable because an unexpected resonance may alter both the amplitude and phase response.

Measure Each Frequency Band Separately

The first useful comparison comes from measuring the low- and high-frequency sections independently. Temporarily disconnect or mute one driver at a time, then capture an impulse response and frequency response for each operating band. Use identical microphone placement, signal level, smoothing, and time reference.

The measurements should overlap through the intended crossover range. Look for the acoustic roll-off slopes, the phase traces, and the arrival time of each driver. A horn compression driver may appear earlier or later than the woofer because its diaphragm sits at a different acoustic depth. The crossover network cannot correct every geometric difference by itself, especially in a passive design.

Near-field measurements can help inspect an individual woofer or horn throat, but they should not be treated as the final phase verdict. Near-field data uses different acoustic references and may not represent the combined radiation pattern at the listening position. Far-field or gated measurements reveal how the drivers actually integrate for the listener.

Check Summation Through the Crossover

With both drivers connected, measure the complete loudspeaker and compare its response with the individual traces. A successful acoustic crossover generally shows a smooth combined curve, predictable directivity, and no unexplained notch at the handoff. The exact target depends on the filter topology, driver behavior, horn coverage, and intended voicing.

A powerful diagnostic is the reverse-polarity test. First measure the loudspeaker with normal polarity. Then reverse the polarity of one driver and measure again without changing the microphone position. If the crossover produces close to a 180-degree phase relationship at the crossover frequency, the reversed-polarity response should show a pronounced, narrow cancellation.

Test What it reveals Typical limitation Useful interpretation
Individual driver sweep Roll-off, phase, and arrival time of each way Does not show acoustic summation Establishes the starting reference
Both drivers in normal polarity Overall response and crossover integration Room reflections can mask small errors Smooth overlap suggests constructive addition
One driver reversed Phase relationship near crossover A shallow null may have several causes A deep null supports correct relative phase
Impedance sweep Network resonances and electrical loading Cannot prove acoustic alignment Finds wiring or component problems
Polar or off-axis sweep Directivity match and spatial consistency Requires more measurements and space Reveals angle-dependent cancellations

A deep reverse-polarity null is useful, but it is not a complete pass/fail test. A null can be weakened by unequal driver levels, different directivity, microphone placement, or insufficient time alignment. Conversely, a strong null at one angle may coexist with poor integration elsewhere. Always return to normal polarity and judge the summed response across several small vertical and horizontal movements.

Interpret Phase, Delay, and Polarity Together

Wrapped phase plots can look confusing because they jump by 360 degrees. Unwrapped phase is easier to interpret over a limited frequency range, but it should be read alongside the impulse response and excess group delay. A sudden phase rotation or delay peak around the crossover may indicate a resonant network, a driver anomaly, or a measurement artifact.

Time alignment is often assessed by comparing the impulse peaks or the acoustic centers of the drivers. However, the highest peak is not always the most meaningful reference. Horns and woofers have different frequency-dependent radiation behavior, so the effective acoustic center changes with frequency. Use the crossover region itself as the priority rather than aligning only one isolated impulse feature.

Polarity must be checked at the terminals and through the complete signal path. A driver can be wired with apparently correct red-to-red connections while its acoustic polarity is reversed by a network topology. Filter order also matters: some electrical slopes introduce phase rotation that changes the required driver polarity. Confirm the schematic, measure the drivers separately, and use the reverse-polarity null as an acoustic cross-check.

Use Listening to Confirm the Measurements

Measurement determines whether the crossover is behaving coherently; listening determines whether the result works as a complete musical instrument. Use familiar recordings with centered vocals, acoustic percussion, bass guitar, and wide-band transients. Listen for a stable center image, even tonal density, and a continuous sense of attack through the crossover range.

Move slightly above and below the tweeter or compression driver axis. A good crossover should maintain a consistent tonal balance over a practical listening window. Large tonal changes or a vocal image that shifts vertically may indicate a directivity mismatch, acoustic offset, or crossover frequency that does not suit the horn and driver combination.

For custom loudspeakers, document every revision. Record component values, driver polarity, microphone distance, crossover frequency, measured null depth, and listening observations. This prevents subjective impressions from being separated from the physical changes that created them and makes it easier to return to a known-good configuration.

Practical Checks Before Finalizing the Network

Use these checks as a compact verification routine:

A passive crossover should be tested at the level where the system will normally operate, then repeated at higher output to identify compression or thermal changes. If the phase relationship shifts substantially with level, investigate driver behavior, saturation, or component limitations before changing filter values.

The final decision should balance electrical behavior, acoustic summation, directivity, and listening. A mathematically elegant phase trace is of limited value if the system has an uneven power response, while a smooth on-axis curve can conceal a narrow listening sweet spot. The strongest design is the one that remains coherent across frequency, level, and realistic listening positions.

When a crossover network is ready for serious evaluation, a controlled listening room and repeatable measurement process can reveal issues that home experimentation may miss. Contact Sunship Audio to discuss a custom horn-loaded system, arrange a demonstration in Berlin, and hear how carefully aligned drivers and passive networks translate into a continuous musical presentation.